We present the results of cognitive tests on conceptual combinations, performed using specific Large Language Models (LLMs) as test subjects. In the first test, performed with ChatGPT and Gemini, we show that Bell's inequalities are significantly violated, which indicates the presence of 'quantum entanglement' in the tested concepts. In the second test, also performed using ChatGPT and Gemini, we instead identify the presence of 'Bose-Einstein statistics', rather than the intuitively expected 'Maxwell-Boltzmann statistics', in the distribution of the words contained in large-size texts. Interestingly, these findings mirror the results previously obtained in both cognitive tests with human participants and information retrieval tests on large corpora. Taken together, they point to the 'systematic emergence of quantum structures in conceptual-linguistic domains', regardless of whether the cognitive agent is human or artificial. Although LLMs are classified as neural networks for historical reasons, we believe that a more essential form of knowledge organization takes place in the distributive semantic structure of vector spaces built on top of the neural network. It is this meaning-bearing structure that lends itself to a phenomenon of evolutionary convergence between human cognition and language, slowly established through biological evolution, and LLM cognition and language, emerging much more rapidly as a result of self-learning and training. We analyze various aspects and examples that contain evidence supporting the above hypothesis. We also advance a unifying framework that explains the pervasive quantum organization of meaning that we identify.

Current AI agents excel in familiar settings, but fail sharply when faced with novel tasks with unseen vocabularies -- a core limitation of procedural memory systems. We present the first benchmark that isolates procedural memory retrieval from task execution, evaluating whether agents can recognize functionally equivalent procedures that span different object instantiations. Using ALFWorld, we construct dual corpora of expert and LLM-generated trajectories and evaluate six retrieval methods using systematically stratified queries. Our results expose a clear generalization cliff: embedding-based methods perform strongly on familiar contexts, yet degrade considerably on novel ones, while LLM-generated procedural abstractions demonstrate reliable cross-context transfer. Controlled ablations show that although embeddings capture some lexical-level abstraction, they fundamentally treat procedures as unordered bags of words, discarding temporal structure necessary for cross-context transfer. Corpus scale delivers far larger gains than representation enrichment, revealing an architectural ceiling in current encoders. Our benchmark offers the first diagnostic framework separating genuine procedural understanding from surface-level memorization and gives tools for developing retrieval systems capable of dependable generalization. Resources available at our GitHub repository (https://github.com/qpiai/Proced_mem_bench).

Building reliable retrieval-augmented generation (RAG) systems requires more than adding powerful components; it requires understanding how they interact. Using ablation studies on 50 queries (15 answerable, 10 edge cases, and 25 adversarial), we show that enhancements such as hybrid retrieval, ensemble verification, and adaptive thresholding provide almost no benefit when used in isolation, yet together achieve a 95% reduction in abstention (from 40% to 2%) without increasing hallucinations. We also identify a measurement challenge: different verification strategies can behave safely but assign inconsistent labels (for example, "abstained" versus "unsupported"), creating apparent hallucination rates that are actually artifacts of labeling. Our results show that synergistic integration matters more than the strength of any single component, that standardized metrics and labels are essential for correctly interpreting performance, and that adaptive calibration is needed to prevent overconfident over-answering even when retrieval quality is high.

How to enable human-like long-term memory in large language models (LLMs) has been a central question for unlocking more general capabilities such as few-shot generalization. Existing memory frameworks and benchmarks focus on finding the optimal memory compression algorithm for higher performance in tasks that require recollection and sometimes further reasoning. However, such efforts have ended up building more human bias into the compression algorithm, through the search for the best prompts and memory architectures that suit specific benchmarks, rather than finding a general solution that would work on other data distributions. On the other hand, goal-directed search on uncompressed information could potentially exhibit superior performance because compression is lossy, and a predefined compression algorithm will not fit all raw data distributions. Here we present SUMER (Search in Uncompressed Memory via Experience Replay), an end-to-end reinforcement learning agent with verifiable reward (RLVR) that learns to use search tools to gather information and answer a target question. On the LoCoMo dataset for long-context conversation understanding, SUMER with Qwen2.5-7B-Instruct learned to use search tools and outperformed all other biased memory compression approaches and also the full-context baseline, reaching SOTA performance (43% gain over the prior best). We demonstrate that a simple search method applied to raw data outperforms goal-agnostic and biased compression algorithms in current long-context memory tasks, arguing for new paradigms and benchmarks that are more dynamic and autonomously scalable. Code for SUMER and all implemented baselines is publicly available at https://github.com/zycyc/SUMER.

Lightweight language models remain attractive for on-device and privacy-sensitive applications, but their responses are highly sensitive to prompt quality. For open-ended generation, non-expert users often lack the knowledge or time to consistently craft high-quality prompts, leading them to rely on prompt optimization tools. However, a key challenge is ensuring the optimized prompts genuinely align with users' original intents and preferences. We introduce PromptTailor, a system for controllable prompt generation for open-ended text that improves model output quality by intent-aligned prompt synthesis. PromptTailor expands minimal user instructions into rich, domain-aware prompts while preserving the user's stated preferences. The system is a quantized Llama3-8B model fine-tuned with a lightweight LoRA adapter on 12,300 prompt-refinement dialogues spanning 41 everyday domains, distilled from three stronger LLMs. The adapter attaches to any Llama3-8B base, enabling edge deployment. In human and LLM-judge evaluations across multiple target models and optimization baselines, PromptTailor yields higher preference rates than chain-of-thought prompting and matches or surpasses state-of-the-art prompt optimization methods while requiring fewer model calls (e.g., 3 vs. 9). These results show that a compact student, guided by powerful teachers, can learn effective prompt-generation strategies that enhance response quality while maintaining alignment with user intent.

The use of Large Language Models (LLMs) for simulating human perspectives via persona prompting is gaining traction in computational social science. However, well-curated, empirically grounded persona collections remain scarce, limiting the accuracy and representativeness of such simulations. Here we introduce the German General Personas (GGP) collection, a comprehensive and representative persona prompt collection built from the German General Social Survey (ALLBUS). The GGP and its persona prompts are designed to be easily plugged into prompts for all types of LLMs and tasks, steering models to generate responses aligned with the underlying German population. We evaluate GGP by prompting various LLMs to simulate survey response distributions across diverse topics, demonstrating that GGP-guided LLMs outperform state-of-the-art classifiers, particularly under data scarcity. Furthermore, we analyze how the representativity and attribute selection within persona prompts affect alignment with population responses. Our findings suggest that GGP provides a potentially valuable resource for research on LLM-based social simulations that enables more systematic explorations of population-aligned persona prompting in NLP and social science research.

Behavioral health conditions, which include mental health and substance use disorders, are the leading disease burden in the United States. Peer-run behavioral health organizations (PROs) critically assist individuals facing these conditions by combining mental health services with assistance for needs such as income, employment, and housing. However, limited funds and staffing make it difficult for PROs to address all service user needs. To assist peer providers at PROs with their day-to-day tasks, we introduce PeerCoPilot, a large language model (LLM)-powered assistant that helps peer providers create wellness plans, construct step-by-step goals, and locate organizational resources to support these goals. PeerCoPilot ensures information reliability through a retrieval-augmented generation pipeline backed by a large database of over 1,300 vetted resources. We conducted human evaluations with 15 peer providers and 6 service users and found that over 90% of users supported using PeerCoPilot. Moreover, we demonstrated that PeerCoPilot provides more reliable and specific information than a baseline LLM. PeerCoPilot is now used by a group of 5-10 peer providers at CSPNJ, a large behavioral health organization serving over 10,000 service users, and we are actively expanding PeerCoPilot's use.

Recent research on large language model (LLM) jailbreaks has primarily focused on techniques that bypass safety mechanisms to elicit overtly harmful outputs. However, such efforts often overlook attacks that exploit the model's capacity for abstract generalization, creating a critical blind spot in current alignment strategies. This gap enables adversaries to induce objectionable content by subtly manipulating the implicit social values embedded in model outputs. In this paper, we introduce MICM, a novel, model-agnostic jailbreak method that targets the aggregate value structure reflected in LLM responses. Drawing on conceptual morphology theory, MICM encodes specific configurations of nuanced concepts into a fixed prompt template through a predefined set of phrases. These phrases act as conceptual triggers, steering model outputs toward a specific value stance without triggering conventional safety filters. We evaluate MICM across five advanced LLMs, including GPT-4o, Deepseek-R1, and Qwen3-8B. Experimental results show that MICM consistently outperforms state-of-the-art jailbreak techniques, achieving high success rates with minimal rejection. Our findings reveal a critical vulnerability in commercial LLMs: their safety mechanisms remain susceptible to covert manipulation of underlying value alignment.

LLMs are sensitive to prompting, with task performance often hinging on subtle, sometimes imperceptible variations in phrasing. As a result, crafting effective prompts manually remains challenging and time-consuming. Recent automatic prompting methods mitigate this difficulty but face three key limitations: (i) for each new task, they require large datasets to train good prompts;(ii) they rely on costly optimization loops that may take hours; (iii)they typically produce a single task-level prompt that does not adapt to the individual input problem to be solved. We propose GPS, the first general-purpose, per-sample prompting method. Without any task-specific tuning, GPS generates a tailored prompt for each unseen input, improving performance across diverse tasks. The prompter is trained with reinforcement learning on a suite of training tasks and includes a novel regularization for effectively adapting to per-sample prompting. Finally, we employ Minimum Bayes Risk decoding to stabilize inference. Empirically, GPS demonstrates competitive performance: we attain second best results among baselines on text simplification, third best results on summarization and on-par results on classification, while not training on any of these tasks, in contrast to the baselines. For in-domain prompting, we obtain sota on GSM8K. Our work shows the potential of a novel and effective paradigm for automatic prompting: generating adaptive, input-specific prompts without extensive optimization and without access to a task-specific training set. Our code is available at https://github.com/Batorskq/GPS.

Environmental, Social, and Governance (ESG) reports have become central to how companies communicate climate risk, social impact, and governance practices, yet they are still published primarily as long, heterogeneous PDF documents. This makes it difficult to systematically answer seemingly simple questions. Existing tools either rely on brittle rule-based extraction or treat ESG reports as generic text, without explicitly modelling the underlying reporting standards. We present \textbf{EulerESG}, an LLM-powered system for automating ESG disclosure analysis with explicit awareness of ESG frameworks. EulerESG combines (i) dual-channel retrieval and LLM-driven disclosure analysis over ESG reports, and (ii) an interactive dashboard and chatbot for exploration, benchmarking, and explanation. Using four globally recognised companies and twelve SASB sub-industries, we show that EulerESG can automatically populate standard-aligned metric tables with high fidelity (up to 0.95 average accuracy) while remaining practical in end-to-end runtime, and we compare several recent LLM models in this setting. The full implementation, together with a demonstration video, is publicly available at https://github.com/UNSW-database/EulerESG.